Analyzing energy consumption and GDP nexus using maximum entropy bootstrap: The case of Turkey
Introduction
Studying the causal relationship between energy consumption (EC) and income is of course nothing new. Since the initial work of Kraft and Kraft (1978), different authors studied this topic and reported contradicting results for different countries as well as for different time periods within the same country.1
Turkey alone has been a subject of at least ten articles published in recent years. Soytas and Sari (2003) used a vector error correction model (VECM) and found that causality runs from EC to GDP for the 1960–1995 period in Turkey. Altinay and Karagol (2004), employing the Hsiao's version of Granger method for the 1950–2000 period, found no evidence of causality between EC and GDP. For the same time period, Altinay and Karagol (2005) used a VAR model along with standard Granger tests and found causality running from electricity consumption to GDP. Jobert and Karanfil (2007) focused on the 1960–2003 period and, based on a cointegration and Granger causality analysis, concluded that no causal relationship exists between GNP and EC in the long run. Halicioglu (2007) employed a VECM approach for the 1968–2005 period and found causality running from GNP to electricity consumption in the long run. Lise and Montfort (2007) undertook an error correction model (ECM) approach for the 1970–2003 period and concluded that causality runs from GDP to EC. Narayan and Prasad (2008) used a basic parametric IID bootstrap approach for studying the OECD countries and found for Turkey no evidence of any causal relationship between GDP and EC between 1960 and 2002. Karanfil (2008), using data for the 1970–2005 period, also concluded that EC and GDP are neutral to each other. Erdal et al. (2008) employed a pair-wise Granger causality analysis for the 1970–2006 period and found bi-directional causality between EC and GNP. Recently, Halicioglu (2009) used an autoregressive distributed lag (ARDL) approach for the 1960–2005 period and found no causal relationship between EC and GNP in Turkey.
Understanding the nature of a possible causal nexus between EC and income has important implications for energy policy in Turkey. Over the last 30 years, Turkey regularly achieved high growth rates while her energy consumption more than tripled during the same period (World Energy Council, Turkish National Committee, 2008). In May 2009, Turkey also ratified the Kyoto Protocol to the United Nations Framework Convention on Climate Change (UNFCCC) and accepted a commitment to plan and reduce greenhouse gas (GHG) emissions starting with 2012. Consequently, if the so-called “growth hypothesis” that EC results in more output is true, energy conservation policies can be detrimental to future economic growth in Turkey. However, if there is a unidirectional causality running from economic growth to EC (“conservation hypothesis”), it may be possible to implement energy conservation policies and cut GHG emissions with little or no adverse effects on output. In fact, a possible negative causality running from output to EC can even result in energy conservation policies increasing the real GDP. On the other hand, neither energy conservation nor expansion policies will have any effect on economic growth if the “neutrality hypothesis” holds, which means that a causal relationship does not exist between EC and GDP.
Despite the potentially crucial policy implications, the inconsistency of the existing findings on the energy–income relationship currently makes it impossible to suggest a reliable policy direction for Turkey. The conflicting results are perhaps not surprising given the evolutionary nature of time series data along with the limited number of available observations. Together, these seem to create empirical results with a high sensitivity to the time period considered as well as the econometric methodology used. In response to the growing number of controversial results, Karanfil, 2009, Ozturk, 2010 argued that future research on this subject should focus on state of the art econometric techniques rather than employing the usual methods for different countries and different intervals of time. We second this proposition and bring into play the maximum entropy bootstrap (meboot) technique. Simulation based hypothesis testing is long known to yield in small samples substantially more accurate results in comparison to conventional inferences based on asymptotic theory. In the energy economics literature, however, bootstrapping has been rarely employed, partly because of the absence of a bootstrap technique useful for strongly dependent time series data.2 The recently developed meboot data generation process (DGP) is specifically designed to fill this gap. It can be employed in all forms of structural breaks and nonstationarity without transforming the data and allows hypothesis testing that is not only accurate, but also robust in the sense of avoiding specification errors. Our objective is to employ this advanced technique to provide conclusive evidence regarding short run precedence also known as Granger causality between energy consumption and GDP in Turkey.
Section snippets
Methodology and the results
When the sample size is relatively small, the traditional hypothesis tests and confidence intervals based on asymptotic theory can yield seriously misleading results. As an example, MacKinnon (2002) discusses how an asymptotic J test at the 5% level can reject a true null hypothesis more than 80% of the time for sample sizes as large as 50. The significance of such over-rejection from the perspective of causality testing is, of course, the risk of wrongly finding a statistically significant
Cointegration analysis
Testing for unit roots and cointegration is not germane to our study because, as discussed earlier, meboot can be seamlessly applied under all sorts of non-stationarity. Still, a formal investigation of the time series properties of the data can be useful to illustrate the advantages of our approach in the analysis of the causal relationship between macroeconomic variables.
Table 2 presents the results for the KPSS, ADF, Engle–Granger, and Johansen–Juselius tests for the different periods as
Conclusion
In the last ten years, Turkey has experienced significant development and has become the 16th largest economy in the world by purchasing power parity (International Monetary Fund, 2010). Due to the growing population and ongoing industrialization, energy investments remain of crucial importance for the country. Turkey is also strategically located at the crossroads of the world's largest oil and natural gas routes, where a number of large multinational energy investment projects are being
Acknowledgments
We would like to thank Fatih Ozatay, the anonymous referees, and the editor Richard Tol for helpful comments and suggestions that have led to significant improvements in the paper.
References (34)
- et al.
Structural break, unit root, and the causality between energy consumption and GDP in Turkey
Energy Economics
(2004) - et al.
Electricity consumption and economic growth: evidence from Turkey
Energy Economics
(2005) - et al.
Economic growth and energy consumption causal nexus viewed through a bootstrap rolling window
Energy Economics
(2010) - et al.
The causality between energy consumption and economic growth in Turkey
Energy Policy
(2008) Some recent developments in a concept of causality
Journal of Econometrics
(1988)Residential electricity demand dynamics in Turkey
Energy Economics
(2007)An econometric study of CO2 emissions, energy consumption, income and foreign trade in Turkey
Energy Policy
(2009)- et al.
Sectoral energy consumption by source and economic growth in Turkey
Energy Policy
(2007) Energy consumption and economic growth revisited: does the size of unrecorded economy matter?
Energy Policy
(2008)How many times again will we examine the energy income nexus using a limited range of traditional econometric tools?
Energy Policy
(2009)
Energy consumption and GDP in Turkey: is there a co-integration relationship?
Energy Economics
Electricity consumption–real GDP causality nexus: evidence from a bootstrapped causality test for 30 OECD countries
Energy Policy
A literature survey on energy–growth nexus
Energy Policy
Energy consumption and GDP: causality relationship in G-7 countries and emerging markets
Energy Economics
Bootstrap methods: applications in econometrics
Ranking mutual funds using unconventional utility theory and stochastic dominance
Journal of Empirical Finance
The size distortion of bootstrap tests
Econometric Theory
Cited by (39)
Forecasting natural gas consumption using Bagging and modified regularization techniques
2022, Energy EconomicsCitation Excerpt :In addition, MEB is of straightforward implementation and is available in different statistical packages2. Hence, the procedure has been effectively applied while investigating associations between energy consumption and economic health in Turkey (Yalta, 2011) and as an auxiliary technique when estimating air temperature quantiles in Central Europe (Barbosa et al., 2011). Notwithstanding these observations, we are unaware of previous applications of MEB in time series forecasting.
A survey of literature on energy consumption and economic growth
2021, Energy ReportsFiscal policy and CO<inf>2</inf> emissions from heterogeneous fuel sources in Thailand: Evidence from multiple structural breaks cointegration test
2020, Science of the Total EnvironmentEncouraging private corporate investment in India
2020, Handbook of Statistics